Back to Docs

Research Papers

Academic research and cutting-edge papers that form the foundation of the Awareness framework

28
Total Papers
9
Conferences
7
Categories
2024-25
Latest Research

🧠Memory & Context Management

🔍Retrieval & RAG

Self-RAG

arXiv:2310.11511ICLR 2024

Topics: Reflection tokens ([Retrieve], [IsRel], [IsSup], [IsUse]), adaptive retrieval

Usage in Awareness: Reflection-based retrieval control

Read full details

CRAG (Corrective RAG)

Topics: Corrective retrieval workflow, decompose-then-recompose algorithm

Usage in Awareness: Retrieval evaluation and correction flow

Topics: Community-based knowledge organization, Leiden algorithm, three query modes (Local, Global, DRIFT)

Usage in Awareness: Knowledge graph construction and querying

LLMLingua-2

ACL 2024

Topics: Binary token classification for fast context compression (3-6x faster)

Usage in Awareness: Context compression technique

LongLLMLingua

Topics: Contrastive perplexity scoring for 'lost in middle' problem

Usage in Awareness: Position-aware context compression

RECOMP

ICLR 2024

Topics: Extractive/abstractive compression for RAG, selective augmentation

Usage in Awareness: RAG-specific compression methods

StreamingLLM

Topics: Attention sinks, KV-cache optimization for long contexts

Usage in Awareness: Context window extension

🎯Planning & Reasoning

LATS (Language Agent Tree Search)

arXiv:2310.04406ICML 2024

Topics: Monte Carlo Tree Search with LLM-based expansion, UCB1 selection, reflection evaluation

Performance: 92.7% on HumanEval

Usage in Awareness: Planning algorithm implementation

Read full details

Tree-of-Thoughts

Topics: Exploration with backtracking, BFS/DFS search strategies

Usage in Awareness: Alternative planning approach

Plan-and-Solve

Topics: Problem decomposition, reduces calculation errors

Usage in Awareness: Planning methodology

🎭Multi-Agent Orchestration

AsyncThink

arXiv:2510.26658

Topics: Fork/Join concurrency, critical path optimization, thinking concurrency reward

Usage in Awareness: Asynchronous agent orchestration foundation

Read full details

Puppeteer

arXiv:2505.19591NeurIPS 2025

Topics: RL-trained orchestration using REINFORCE, dynamic agent selection

Usage in Awareness: Learned orchestration policy

AFLOW

Topics: MCTS workflow optimization, automated workflow discovery

Usage in Awareness: Workflow discovery and optimization

Optima

Topics: Communication efficiency training, reward function (accuracy + efficiency + readability)

Performance: 2.8x performance with <10% tokens

Usage in Awareness: Multi-agent communication efficiency

ADAS (Adaptive Dynamic Agent Selection)

Topics: Meta agent search for workflow discovery

Usage in Awareness: Agent selection strategies

DyLAN

Topics: Dynamic agent selection and pruning, importance scoring with backward propagation

Usage in Awareness: Team selection methodology

Topics: Multi-agent reinforcement learning for LLM collaboration

Usage in Awareness: Multi-agent training approach

🛠️Tools & Capabilities

Toolformer

Meta, 2023

Topics: Self-supervised tool learning, when/how to call tools

Usage in Awareness: Tool usage training

OpenHands SDK

Topics: Production coding agent architecture

Usage in Awareness: Production-ready agent patterns

📏Context Extension & Optimization

LongRoPE

ICML 2024

Topics: Progressive RoPE rescaling, 2M+ token context

Usage in Awareness: Position interpolation for context extension

YaRN

Topics: Non-uniform interpolation across RoPE dimensions

Usage in Awareness: Alternative context extension approach

KVQuant

NeurIPS 2024

Topics: Sub-4-bit KV cache quantization, 10M context on A100-80GB

Usage in Awareness: KV-cache optimization

🔗Vector Search & Indexing

HNSW (Hierarchical Navigable Small World)

Topics: Graph-based vector indexing, O(log n) search complexity

Usage in Awareness: Vector search algorithm foundation